Explorations 2.0¶
Since the release of Explorations, our AI-powered agent that helps you explore your data in the UI, we've been gathering feedback. Now, we are releasing a major revamp.
It now includes three modes:
- Chat mode: Focus on text output with a cleaner interface, logic organized into collapsible steps, suggested follow-up questions, and more.
- Free query mode: Write your own SQL queries. A chat assistant is available in fullscreen mode to help you.
- Time series mode: Specifically designed for data to review over time. You can configure them manually by specifying the Data Source, time field, and metric to observe, or you can ask an LLM.
See the Explorations documentation. Try it out and give us your feedback through the community.
Track your LLM usage across Tinybird¶
New Service Data Sources are now available to track LLM usage in your Workspace and Organization.
tinybird.llm_usageincludes data for your Workspace.organization.llm_usageprovides the same data at the Organization level and is only available to Organization admins.
These Data Sources include all requests sent to the LLM features in Tinybird, such as Explorations and Tinybird Code. Each request includes cost, tokens used, duration, and more.
Organizations on Free Plans have limited access to LLM features with 25 credits per month.
New system tables available as Service Data Sources¶
Three new system tables are now available as Service Data Sources in Tinybird:
These Data Sources provide access to internal system information that can help you understand operational details beyond what's available in existing Service Data Sources like pipe_stats_rt or datasources_ops_log.
While some of this information may overlap with existing Service Data Sources, these system Data Sources offer additional internal details that can be valuable for advanced monitoring, debugging, and analytics use cases.
These system Data Sources are not as optimized as standard Tinybird Service Data Sources and can be very resource-intensive. They may hit query limits more easily than other Data Sources. Always filter by event_time when querying these Data Sources to ensure optimal performance and avoid timeouts.
New metrics added to metrics_logs Service Data Sources¶
Three new metrics are now available in the metrics_logs Service Data Sources for enhanced monitoring of replication:
- ReplicasSumQueueSize: Sum queue size (in the number of operations like get, merge) across Replicated tables
- ReplicasSumMergesInQueue: Sum of merge operations in the queue (still to be applied) across Replicated tables
- ReplicasSumInsertsInQueue: Sum of insert operations in the queue (still to be applied) across Replicated tables
These metrics provide better visibility into the operational status and performance of Replicated tables in your ClickHouse clusters. For more information about the metrics_logs Service Data Sources, see the monitoring documentation.
Events API Request Size Limit Will Be Enforced for Developer and Enterprise Plans Starting December 1, 2025¶
To ensure consistent high performance and reliability for all Events API users, we will begin enforcing the maximum request size limit on Developer Plans, starting December 1, 2025. As noted in the Events API documentation, the maximum payload size for Developer and Enterprise Plans is 100 MB. This limit has already been enforced on Free Plans for payloads larger than 10 MB, and will also be applied to Developer and Enterprise Plans starting December 1.
If your requests exceed 100 MB, please split them into smaller batches to avoid rejection. Starting December 1, 2025, requests over the limit will be rejected with an HTTP 413 - Request Entity Too Large error.